Visual attention and spatial
working memory. Several current research
studies are investigating how infants, children, and adults
use both memory and visual skill to guide their
attention across space and time. I also study how
people
divide
their
attention
while working on two or more simultaneous
tasks. As noted below, this work is complemented by a number
of computer
models that simulate eye-movement activity in real time.
Neural
network models of vison and oculomotor control. In
collaboration with Dima
Amso (Cornell University)
and Scott Johnson (UCLA),
I also design and test computer models of early visual
processing and oculomotor control. These
models have three important features:
- Biological
inspiration: the architectures and learning
algorithms used in these models are inspired by the anatomy
and physiology
of the mammalian visual system
- Embodied
knowledge:
these models are designed with simple physical bodies,
which enable them to interact with and learn from their
environment--a
critical element of robotics and machine learning models
that must deal with noisy or uncertain environments
- Realtime
performance: although current modeling work is computer-simulated,
the goal of each model is to reproduce behavior (e.g., eye
movements, hand and arm movements) as it unfolds in real time
Developmental
cognitive neuroscience. A third project
uses neural-imaging methods (e.g., fMRI) to investigate the
neural substrates of working memory and spatial-directed
attention. This work not only informs the computer models that
are being developed, but it is also shaped and influenced by
the results of the modeling work.
